Summary
This paper presents a stochastic framework to properly involve the uncertainties associated with demand in distribution system restoration (DSR) problem. To reach this goal, these uncertainties are represented as probabilistic scenarios corresponding to different load levels. Subsequently, the associated stochastic optimization problem is formulated such that it can be readily solved using dynamic programming approach. Moreover, a clustering technique is presented that enables the dynamic programming approach to find near‐optimal solutions for the stochastic load restoration problem with reasonable computational effort. The proposed framework is implemented on a real‐world test system, and its efficiency and effectiveness are demonstrated through extensive case studies. The results suggest that incorporation of load uncertainties in the DSR problem would significantly affect the optimal sequence of feeder restoration, and therefore, appropriate treatment of load uncertainties in DSR problem is essential.